XiKey™DESKTOP
العربيةGet XiLauncher

Creative Studio

ALPHA · FREE

XiArt

Shape the scene. Then explore the possibilities.

XiArt brings the ArtCraft approach to AI image and video creation into the XiKey™ family: compose a canvas, stage a scene and guide generation visually. This is a free Alpha integration project; upstream capabilities are documented below, while XiKey packages and hosted model access remain unverified.

Alpha is free today.

Shape the scene. Then explore the possibilities.

Explore the details

Compose: place the elements where you want them
01

Compose: place the elements where you want them

Arrange image layers and drawing on a 2D canvas before asking a model to refine the composition. This is an implemented ArtCraft workflow; XiArt integration is Alpha work.

Source ↗

Checked: 2026-10-14

02

Compose: change a selected area

Use brushes and masks to guide inpainting instead of regenerating a whole image. A compatible generation service is required.

Source ↗

Checked: 2026-10-14

03

Compose: reuse a subject without its backdrop

The upstream background-removal workflow prepares image cutouts for canvas layers and scene props; processing depends on its provider.

Source ↗

Checked: 2026-10-14

Direct: give flat images a sense of depth
04

Direct: give flat images a sense of depth

Stage backdrops, foreground images and props in 3D so you can decide their spatial relationship before generation.

Source ↗

Checked: 2026-10-14

05

Direct: block a scene with reusable pieces

Combine 3D assets to test object placement, depth and camera angles rather than describing all of them only in a prompt.

Source ↗

Checked: 2026-10-14

06

Direct: choose the pose before the shot

The upstream posing workflow lets you arrange a character and camera to guide the generated shot. It does not guarantee exact anatomy or identity.

Source ↗

Checked: 2026-10-14

07

Direct: guide a character with a posed reference

Use the documented identity-transfer workflow to combine a character reference with mannequin placement and pose; results vary with the model.

Source ↗

Checked: 2026-10-14

08

Direct: explore shots in a shared environment

The Image to Location workflow builds a spatial setting from an image for virtual actors and multiple viewpoints, subject to service availability.

Source ↗

Checked: 2026-10-14

09

Direct: turn a reference into a movable 3D prop

Request a 3D mesh from an image, then position and rotate it in the scene. Mesh generation is provider-dependent.

Source ↗

Checked: 2026-10-14

Generate: explore a visual idea from words
10

Generate: explore a visual idea from words

The upstream prompt workflow offers image-model selection and text-to-image requests. XiKey has not verified live model access or included generation credits.

Source ↗

Checked: 2026-10-14

11

Generate: ask for a targeted visual revision

Describe a change to an existing image through a supported editing model. Which models work depends on the connected service and account.

Source ↗

Checked: 2026-10-14

12

Generate: explore movement from a still frame

Send a starting image to the upstream video-generation workflow and choose available controls. This is not a promise of an offline renderer or a free video service.

Source ↗

Checked: 2026-10-14

Give an idea something you can direct

A prompt is a useful beginning. Sometimes the next decision is visual: move the subject to the edge, lower the camera, leave room for a headline, or make the foreground carry more weight. XiArt takes its foundation from ArtCraft, which combines a layered 2D canvas with a 3D staging space and provider-connected image and video generation. You can develop the composition before asking a model to interpret it. These are documented upstream workflows; XiArt's own integration and service access remain under development. Upstream capabilities.

XiArt is for illustrators, art directors and small creative teams exploring alternatives before committing to a finished asset. A promising generated image still needs your judgment about composition, factual details and whether it matches the brief. No claim of exact character consistency, superior image quality or guaranteed model access is made here.

Three briefs, three different ways to begin

A puppet theatre's next stage. Build a rough scene from backdrop images and movable props, then test a low camera against a higher view. Use a posed reference to explore the relationship between performer and scenery. The purpose is a discussion board for the director, rather than a construction drawing or a guarantee of the final staging.

A botanical garden's pollinator panel. Start with licensed photographs, separate a subject from its background and assemble the surrounding flowers as layers. Mask the area that needs an illustrative transition. Review plant and insect details with the garden's educators before the image becomes public interpretation material.

A luggage designer's concept review. Place a reference object in a spatial scene and compare a clean silhouette with a busier travel backdrop. Try a prompted revision to the visual treatment. If video services are available, explore movement from the chosen still. This is a concept workflow, not dimensional product validation.

These are suggested creative exercises, not bundled sample files or evidence that a XiArt release has passed them. The underlying composition, posing, background-removal and generation workflows are described in the checked ArtCraft README.

A workflow that starts with the composition

  1. Set the brief. Decide what the image must communicate and gather references you have permission to use.
  2. Build the visual guide. Arrange layers and drawing on the 2D canvas, or stage images and meshes in 3D. Choose the pose and viewpoint before requesting a render.
  3. Explore through a connected model. Use a text prompt, a guided image edit or an image-to-video request when a compatible service is available. Model availability and charges belong to that service.
  4. Refine and inspect. Change an area with a mask, revise the composition and review the result. The upstream scene editor can download a PNG snapshot; generation outputs and supported controls vary by provider.

Evidence: canvas implementation, staging workflow, PNG snapshot export.

What XiKey™ is adding

Arabic belongs in the working interface. An Arabic UI and one-click EN ⇄ ع switch are in progress for XiArt. The switch already exists in the XiPhoto pilot; it is not a shipped XiArt feature. The Arabic font dialog, grouped by writing style, and XiKey splash, About and dialogs are also in progress. Shared kashida, tashkeel, Arabic digit and Hijri-date tools are being developed for image/text workflows; XiArt integration remains unfinished.

Classic and Modern layouts with Retina scaling are planned. Sandboxed add-ons managed through XiLauncher are planned, while the signed update channel is in progress. XiArt uses a different upstream toolkit from the egui products, so a shared plan does not prove that their add-on mechanism already works here. Status source: XiKey status, checked 2026-10-14.

Alpha: a clear invitation to explore

XiKey Alpha is free today. XiArt is a planned product being integrated into that Alpha program; no downloadable XiKey package, minimum hardware specification or authenticated live generation route was verified for this page. Upstream documents Windows and macOS downloads and Linux source development, which does not establish XiKey installer availability. Platform evidence, source setup.

The app's announced Beta price is $19/year or $2.49/month. Cloud generation, if offered, is billed separately as credits. App pricing comes from XiKey pricing, checked 2026-10-14; it is not a promise of free inference or access to every upstream model. The upstream catalog itself labels some entries disabled or restricted, and removing dependence on hosted services remains an upstream roadmap item.

What we checked

Research date: 2026-10-14. ArtCraft source configuration reports 0.42.0, at commit d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373. The latest published release checked was artcraft-v0.41.0, published September 26, 2026; its release note only points to downloadable assets. Feature evidence comes from the pinned README, documentation and source rather than treating the unreleased source version as a XiKey release. Code licences were checked in LICENSE-MIT, LICENSE-APACHE and NOTICE. No provider logos or bundled third-party sample assets are used on this page.

XiArt is based on ArtCraft by the ArtCraft team (MIT OR Apache-2.0).

XiKey is a trademark of Pioneer Local Development Co. Ltd.

The XiKey additions

In progress

Arabic interface and one-click EN ⇄ ع

In progress for XiArt. The live language switch is shipped in the XiPhoto pilot; it is not yet a delivered XiArt feature.

In progress

Find Arabic fonts by writing style

The shared font dialog is being developed with Arabic style groups and script-aware previews; XiArt integration is unfinished.

In progress

A consistent XiKey™ welcome and About experience

XiKey splash, About and dialogs are in progress for XiArt, with upstream attribution and licence information retained.

Planned

Classic or Modern layouts, with Retina scaling

Planned interface presets and high-density scaling; not shipped in XiArt.

In progress

Arabic text controls for image compositions

Kashida, tashkeel controls, Arabic digits and Hijri dates are in progress in the shared toolkit; their integration into XiArt image workflows remains unfinished.

Planned

Add-ons managed through XiLauncher

Sandboxed XiKey add-ons through XiLauncher are planned. This does not assert that ArtCraft currently exposes the same plug-in system as the egui applications.

In progress

Signed updates through XiLauncher

The signed update channel is in progress; XiArt delivery and installer validation have not been verified.

System requirements

  • XiArt installer availability and minimum RAM/GPU/OS versions have not been verified; no packaged platforms are asserted here.
  • Upstream documents Windows and macOS downloads and Linux source development; this is not XiKey package availability. Evidence: https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/README.md and https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/_docs/dev_setup.md (checked 2026-10-14).
  • 3D previews use WebGL2; this is a code requirement, not a tested hardware minimum. Evidence: https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/frontend/libs/components/upload-modal/src/lib/UploadModal3D/utilities/loadPreviewOnCanvas.ts.
  • Online generation requires compatible services/accounts; removing hosted-service dependence is still an upstream roadmap item. Evidence: https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/README.md and https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/ROADMAP.md.

File formats

Open

pngjpgjpeggifglbspz

Save

png

A plan for your work

Alpha is free today.

Beta pricing preview

Learn

Open-source foundation

XiArt is based on ArtCraft by the ArtCraft team (MIT OR Apache-2.0).

View upstream repository ↗

MIT OR Apache-2.0 · 0.42.0 (source configuration); commit d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373; latest published release artcraft-v0.41.0 · 2026-10-14

App plan

XiArt

$9.50/ first year

Then $19.00 / year

Save $10.88 versus 12 regular monthly payments

Cloud AI generation, if offered, is billed as separate credits.

All updates included while subscribed, delivered and verified through XiLauncher.

Preview checkout

Questions, answered

Is XiArt free?

The XiKey Alpha policy is free. The announced Beta app subscription is $19/year or $2.49/month; cloud AI generation, if offered, is billed separately as credits. A free app does not mean free inference. XiKey pricing: /store/ (2026-10-14).

Can I use every model listed upstream?

No such guarantee is made. The upstream catalog includes disabled and restricted entries, and the XiKey service integration has not been verified. Evidence: https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/README.md.

Which file formats are evidenced?

The listed open formats are handled in the upstream 3D upload preview, not a claim of universal project interchange: https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/frontend/libs/components/upload-modal/src/lib/UploadModal3D/utilities/loadPreviewOnCanvas.ts. PNG scene snapshots are implemented in https://github.com/storytold/artcraft/blob/d4e4ac69bbd93d237eaf5d0260bc6a5ff3a5e373/frontend/libs/components/pagescene/src/lib/engine/editor.ts. XiArt round-trip tests have not been run.

Is the Arabic interface already available in XiArt?

It is in progress. XiPhoto is the shipped bilingual pilot; the XiArt port, font dialog and text tools remain unfinished. XiKey status: /products/xiart/ (checked 2026-10-14).

Compare

See both sides.

Explore a fair comparison, including where the other product is stronger.

ALPHA · FREE · Checked: 2026-10-14

XiArt / Midjourney

Official pricing: $10.00 / month · Basic, monthly billing, USD; taxes not included in this comparison · Region: US · 2026-10-14 Source ↗

TopicXiArtMidjourney
App cost and generation budget
Source 1 ↗Source 2 ↗
Alpha is free; announced Beta app pricing is $19/year or $2.49/month. Cloud generation, if offered, uses separate credits.Basic costs US$10/month on monthly billing and includes 3.3 Fast GPU hours; it is a generation-service subscription.
From a prompt to a first image
Source 1 ↗Source 2 ↗
ArtCraft implements prompt-driven image generation with model selection; XiKey live service access is unverified.Midjourney has a documented prompt workflow for image and video generation.
Editing the composition
Source 1 ↗Source 2 ↗
Upstream provides a layered 2D canvas, drawing and mask-guided edits; connected inference is required for generated edits.Its web Editor supports layers, move/resize, smart selection and inpainting/outpainting; it is not limited to text prompts.
Spatial direction and reference guidance
Source 1 ↗Source 2 ↗
ArtCraft documents 3D staging, camera placement and posed-character guidance. No benchmark of reference fidelity is claimed.Style References guide a consistent visual treatment; this is a different workflow from positioning objects in a 3D scene.
Animating a starting image
Source 1 ↗Source 2 ↗
Upstream has image-to-video requests across supported providers; XiKey access, limits and costs are not verified.Midjourney generates an initial five-second video from an image; resolution and processing modes depend on plan.
Privacy and service readiness
Source 1 ↗Source 2 ↗
XiArt is Alpha integration work. A local desktop interface is not evidence of private or offline cloud generation; provider terms still matter.Stealth is a documented option on Pro and Mega. Its availability should not be inferred for Basic.